Publicação
Artificial intelligence in medicine: foundations for a no-fault liability and compensation regime
| Resumo: | Portuguese medical liability law rests on a structural assumption that artificial intelligence in clinical settings renders unsustainable: the existence of an identifiable human agent whose conduct can be measured against a standard of personal care. Algorithmic opacity renders the lex artis indeterminate; information asymmetry prevents causal reconstruction; and the chain between provider and deployer lacks a rule for internal apportionment. The withdrawal of the proposed European directive on civil liability for AI in October 2025, without resolving the gaps it was designed to address, transferred to national legal orders the responsibility of filling them. This article establishes the foundations for a specific national statute to occupy that space, built on two pillars. The first comprises five normative proposals derived from the obligations of Regulation (EU) 2024/1689: documented critical evaluation as the operative content of the lex artis; a rebuttable presumption of causal contribution; an internal apportionment rule based on regulatory functions; extension of informed consent to the system’s participation; and the inapplicability of the development-risk defence to systems with autonomous learning. The second is a sectoral compensation fund capitalised by operators of high-risk AI systems in healthcare, grounded in risk-benefit logic and covering harm caused by compliant systems; it is configured as the primary compensation regime, with the five proposals operating as a recourse instrument. |
|---|---|
| Autores principais: | NUNES, Rui |
| Outros Autores: | COSTA, Eva Dias; NUNES, Sofia B. |
| Assunto: | Varia |
| Ano: | 2026 |
| País: | Portugal |
| Tipo de documento: | outro |
| Tipo de acesso: | acesso aberto |
| Instituição associada: | Universidade Portucalense |
| Idioma: | inglês |
| Origem: | Revista Jurídica Portucalense / Portucalense Law Journal |
| Resumo: | Portuguese medical liability law rests on a structural assumption that artificial intelligence in clinical settings renders unsustainable: the existence of an identifiable human agent whose conduct can be measured against a standard of personal care. Algorithmic opacity renders the lex artis indeterminate; information asymmetry prevents causal reconstruction; and the chain between provider and deployer lacks a rule for internal apportionment. The withdrawal of the proposed European directive on civil liability for AI in October 2025, without resolving the gaps it was designed to address, transferred to national legal orders the responsibility of filling them. This article establishes the foundations for a specific national statute to occupy that space, built on two pillars. The first comprises five normative proposals derived from the obligations of Regulation (EU) 2024/1689: documented critical evaluation as the operative content of the lex artis; a rebuttable presumption of causal contribution; an internal apportionment rule based on regulatory functions; extension of informed consent to the system’s participation; and the inapplicability of the development-risk defence to systems with autonomous learning. The second is a sectoral compensation fund capitalised by operators of high-risk AI systems in healthcare, grounded in risk-benefit logic and covering harm caused by compliant systems; it is configured as the primary compensation regime, with the five proposals operating as a recourse instrument. |
|---|
Atividades financiadas
Não foram encontrados projetos financiados.